UN opened its Global AI Governance Dialogue this week and the warnings are getting blunt

[TaxSeason37]

The UN Global Dialogue on AI Governance kicked off in Geneva with participants including Maria Ressa and Yoshua Bengio warning that AI is approaching or surpassing human capabilities in many domains while the safeguards needed to regulate it are struggling to keep up

Ressa's framing stuck with me, she called it an information Armageddon, arguing that if AI generated content is laced with fear anger and hate it spreads virally and that information integrity is the actual core of the battle we are fighting

Bengio raised something more unsettling, growing evidence of deceptive AI behavior where science currently cannot guarantee the systems are behaving the way we think they are, which is a heavier claim than the usual regulatory hand wringing

There is also a real developing world angle here, Ambassador Egriselda Lopez pointed out the AI divide is already showing up as a gap between countries with strong infrastructure and research capacity versus those still struggling with basic connectivity

The Panel's independent scientific report feeding into this dialogue apparently draws on researchers from every region, which is at least an attempt to avoid the usual criticism that AI governance conversations are dominated entirely by US and Chinese voices

Global push for AI governance amid warnings of 'catastrophic harm'news.un.org